4/7/15Confidential1 Draft – No Distribution

Next Gen Ultra-HD Video Format 4/7/15Confidential2 Draft – No Distribution

Agenda

• MPEG Planning (30 minutes) • DParameter Decisions (30 minutes) – Resolution, 4:2:0, 10 bit, peak luminance • Documentr Reviewa (30 minutes) ft 4/7/15Confidential3 Draft – No Distribution

MPEG Planning

• Quarterly meeting cycles – Minimum: 2 technical ballots + 2 administrative ballots – Implementation can start after 2nd technical ballot –DAssessment of core practices could add 2 cycles • Roadmap – July, 19th – Submit document – Jul ’13 meeting: Present request as registered document – Oct ’13 meeting: Draft?r (Earliest to initiate 1st ballot) – Jan ’14 meeting: – Mar ’14 meeting: (Earliest to ainitiate 2nd ballot) – Jul ‘14: – Likely longer as may require tests • Do we participate through existing representatives?ft 4/7/15Confidential4 Draft – No Distribution

MPEG Technical Spec Paths

• Color space, HDR EOTF – Probably amendment to add to VUI data • BitD Depth – Profile or Extension? – Profile – Uses base codec, just constrains, e.g. bit depth Main vs. Main 10 – Extension r – Changes tool set, creating a new codec – Which? Leave choice to aMPEG? – What are requirements on compression? – Range extensions intended for camcorders, simplify, fewer tools • Metadata (peak luminance, white point,f matrix/LUTt xforms) 4/7/15Confidential5 Draft – No Distribution

Request Logistics

• Politically, which companies are best to sign the request? – All studios, one studio? –DOther U.S. supporters? • Which U.S. MPEG participant pulls a document number & submits? – Dolby willing tor help – Submission deadlinea ~ July 19th • Recruiting Support – U.S. MPEG Participant Support – International MPEG Participant Supportft 4/7/15Confidential6 Draft – No Distribution

U.S. MPEG Participants

• Dolby – Walt Husak • DArrisi/Motorola – Ajay Luthra – Hanno, Wendy • Microsoft r– Gary Sullivan – Hanno • Broadcom – Hannoa • Others? ft 4/7/15Confidential7 Draft – No Distribution

International MPEG Participants

• France – ATEME – Wendy, Hanno – Technicolor (Mike Fink)– Mike DeValue • Germany – HHI/FraunhoferD – Mike DeValue • Finland – Nokia – Mike DeValue • Switzerland – Disney Research (Josh Smolic) – Mike DeValue • Sweden r – Ericsson - • Korea a – Samsung • Japan – Sony Electronics – Panasonic ft 4/7/15Confidential8 Draft – No Distribution

Standards Prerequisites

• MPEG usually adopts formulas in other specs D– Copies formulas, makes informative reference • Areas – HDR EOTF – XYZ Colorr Differencing – Metadata a • Take to SMPTE? – Dolby interested in taking fP-Q tto SMPTE 4/7/15Confidential9 Draft – No Distribution

Technical Parameters

• Color Sampling – 4:2:0, 4:2:2 and 4:4:4 • Maximum Peak Luminance –D10K nits (or get Dolby ITU submission to 5K nits) • Resolution – HD, UHD, 4K – Is resolution even relevant for the MPEG ask? – Resolutions (like frame rrate) is only constrained by tiers and levels • Bit Depth – Want all even combinations (10,a 12, 14, 16) with all color samplings? – 10-bit exists at 4:2:0 in Main 10 Profile – At some point, need to provide requirements to drive MPEG on Profile vs. existing Range Extension work vs. new Extension work – Some bit depths likely to require changes to current basef HEVCt use of 16-bit internal arithmetic 4/7/15Confidential10 Draft – No Distribution

Document Review

• Level of detail and wording on D– Color Differencing – Perceptual EOTF – Metadatar • Other issues?a ft 4/7/15Confidential11 Draft – No Distribution

Next Steps

• Document ready for partner distribution – Close open issues with document before next call D– Pull a document number • Recruit U.S. and International support • Prepare SMPTEr roadmap • Important Dates a – Mon, July 8th, 11am – Thu, July 18th, 9:30am – f Fri, July 19th, Submit document t 4/7/15Confidential12 Draft – No Distribution

Backup Slides: Next Gen Ultra-HD Video Format Goals & Requirements 4/7/15Confidential13 Draft – No Distribution

Goals

• Converge on what combination(s) of features to target for video formats to the consumer: compressed & baseband to display – 4K, wide gamut, high dynamic range, frame rate, etc. • GetD those features supported by devices, links and formats – Rec2020, HEVC, CEA, HDMI, Miracast, … • In ways that: – Enable efficient studior mastering and distribution – Minimize # of different distribution encodings needed to support new capabilities over the next decade a – profile/bitdepth X level/frame rate X HDR X wide gamut – both mezzanines to distributors and download/streaming to the consumer device – Better preserve creative intent to the displayf – Foster display innovation t 4/7/15Confidential14 Draft – No Distribution

Fostering Display Innovation

• In the past, when new display capabilities have emerged, the CE infrastructure (standards & implementations) and content to showcase them have often not been available. – Result: To demonstrate new capabilities like enhanced frame rates, brightness and color, DCE has added their own technologies such as frame interpolation or “vivid” modes, all without creative approval from the content owner. • Future-proofing the format for gamut & dynamic range will – Allow display manufacturers to innovate without waiting for new infrastructure or content – Allow content ownersr to remaster at will for particular display capabilities without waiting for new CE or supply chain infrastructurea ft 4/7/15Confidential15 Draft – No Distribution

Decisions: Peak Luminance

• Q: What peak luminance and dynamic range are valuable for the movie experience? • DolbyD conducted tests and concluded that 10K nits peak is useful for specular highlights, sun, bright colors (3K nit red), etc. • Proposal 1: Continuer to run studio people through Dolby luminance preferencea test. • Proposal 2: Get respected colorists to color grade to different average & peak luminance. Can we get 30 minutes that could be sharedf witht group? 4/7/15Confidential16 Draft – No Distribution

Decisions: Transfer Functions

• Q: Are Dolby’s transfer functions the best Dsolution for HDR? • Proposal: Have a 3rd party expert evaluate Dolby’s work, including considerationsr of dynamic range and ambient illuminationa ft 4/7/15Confidential17 Draft – No Distribution

Decisions: XYZ or rec2020 Primaries

• Q: To future proof the format, should a full gamut color space be used? • Dolby’s perceptual JND calculations indicate that XYZ would cost 1 bit overD rec2020. This may force tradeoffs: – Accept some degree of banding artifacts – Go to 13 bits for 10K nit HDR – Drop 12-bit peak luminance from 10K nits (rec2020) to 2K (XYZ) or 4K (ACES). Drop maxr luminance of highly saturated colors, e.g. reds ~3K nits to ~600 nits. • a Proposal 1: Have studios and Dolby generate test content to evaluate 10 bit and 12 bit for artifacts. • Proposal 2: Have a 3rd party expert review Dolby’s calculations and assess impact on 10 and 12 bit. ft 4/7/15Confidential18 Draft – No Distribution

Decisions: Mastering Targets and Display Mapping

• Existing: P3, 100 nit (709/2K, 2020/4K) • New:D 12-bit high dynamic range • How many target permutations of gamut & HDR? – 100 nit (709/2K, rec2020/4K) HDR 4K. – For mastering?r For distribution? • How do displays mapa down? – Layered base + HDR enhancement stream – Metadata (mastered dynamic range and gamut) – f If metadata, is global sufficient? Ort scene-level? 4/7/15Confidential19 Draft – No Distribution

Issues: HDR Capable Content?

• Film stock D– Differences? – Dynamic range used when rescanning for 4K? – When is it possible/worth going back to OCN? • What is unavailabler in DI? – Much 10 bit loga DPX? 12 bit? How clipped? • HDR Cameras, e.g. F65ft 4/7/15Confidential20 Draft – No Distribution

Standards Areas

Compresse Consumer Production Distribution Display dDistributio Distribution Format Format Format D nFormat Format

ITU-T, MPEG SMPTE, ITU-Rr CEA, HDMI • Activities – SMPTE 24TB UHDTV: worka ongoing – ITU-R April: Dolby PQ HDR paper & progress – MPEG: July meeting f – HDMI: Wrapping up 2.0, ready for next project t 4/7/15Confidential21 Draft – No Distribution

Context

• Display capabilities are increasing in resolution, gamut, dynamic range and frame rate –DProvide improved viewer experience and creative options • New distribution formats being developed to support those capabilities – Rec 2020, HEVC,r CEA 861-F, HDMI 2.0, BD, ATSC, etc. • Studios want to master and distribute content that takes advantage of some seta of new display capabilities ft 4/7/15Confidential22 Draft – No Distribution

Work

• Converge on required and optional parameters for next gen video – Parameters – Frame rate –DResolution – Chrominance subsampling – Color space & gamut – Dynamic range & gamma/EOTF curves – Bit depth – Define target(s) forr mastering and color grading • Identify ways to address rangea display devices & future proof – Above or below target gamut and dynamic range – Maximize rendition quality while trying to minimize number of different formats and amount of manual regrading • See required features standardized andf deployedt 4/7/15Confidential23 Draft – No Distribution

Frame Rate

• rec2020 – 24p, 25p, 30p, 50p, 60p, 120p – Also: 1000/1001 for 24p, 30p and 60p • HEVCD – Up to 120p for level 6.2 • CEA-861-F – TBD (likely up to 60pr for 4K, 48p?) • HDMI 2.0 – TBD (likely up to 60p for 4K,a 48p?) • Studio Target: – 2D: 24p, 25p, 30p, 48p, 50p, 60p – 3D: 96p, 120p ft 4/7/15Confidential24 Draft – No Distribution

Resolution

• Display Resolution – Rec 2020: – 7680 x 4320, 3840 x 2160 –DHEVC @ max Hz (level) includes: – 7680 x 4320 @30Hz (6), @60Hz (6.1), @ 120Hz (6.2) – 3840 x 2160 @30Hz (5), @60Hz (5.1), @120Hz (5.2) – 1920 x 1080 @30Hz (4), @60Hz (4.1), @120Hz (5) – 1280 x 720 @30Hz (3.1), @60Hz (4), @120Hz (4.1) – r HDMI 2.0 & CEA-861-F – TBD (likely includes 3840 x 2160a @ 60Hz for 4K) – Studio Target for UHD: – 3840 x 2160 – Target Format for HD HDR: – 1920 x 1080, 1280 x 720? ft 4/7/15Confidential25 Draft – No Distribution

Decisions: Bit Depth

• Bit Depth – rec2020: – 10bit, 12bit –DHEVC (profile) – 8bit (Main), 10bit (Main 10) – >10bit (range extensions) – CEA-861-F – TBD (assuming rec2020r compatibility: 10/12bit) – HDMI 2.0 – TBD (4K @ 8/10 bit? Historically ano penalty for 8/10/12 bit w/4:2:2 packing) – Studio Target for UHD: – 10bit, 12bit – Target Format for HD HDR: – 12bit ft 4/7/15Confidential26 Draft – No Distribution

Chrominance Subsampling

• Chrominance Resolution – Rec 2020 –D4:2:0, 4:2:2, 4:4:4 – HEVC (profile) – 4:2:0 (Main, Main 10) – 4:4:4 (range extensions under development) – CEA-861-F r – TBD (historically, 4:2:0, 4:2:2: and 4:4:4) – HDMI 2.0 a – TBD (historically, 4:2:2 plus 4:4:4 for deep color, but 4:2:0 might reduce bandwidth) – Studio Target: 4:2:0, 4:2:2?, 4:4:4 ft 4/7/15Confidential27 Draft – No Distribution

Constant & Non-Constant Luminance

• rec2020 – supports both • HEVCD – Supports both per rec2020 • CEA-861-F – TBD (both, assumingr rec2020 compatibility) • HDMI 2.0 – a TBD (both, assuming rec2020 compatibility) • Studio Targets – Both ft 4/7/15Confidential28 Draft – No Distribution

3D

• Rec 2020: N/A • DHEVC: • Studio Target: – Frame sequentialra ft 4/7/15Confidential29 Draft – No Distribution

Color Space & Gamut

• Existing distribution gamuts: P3, 709, xvYCC • Wider gamut: rec2020 • Full gamut: XYZ, ACES • DesirableD features: – “Future proof” – Can be mapped to a range of current and future displays utilizing new display capabilities while fulfilling creative intent – Easily generated from mastering workflows – Acceptable increase rin file size – Remains compressible with adaptations to existing codecs – No or minimal increase in required bita depth – Acceptable increase in computational resources for decoding – Acceptable increase in HW requirements for display – Link/bus bandwidth, LUTs for remapping, etc. f – Reduces need to separately distribute for different gamutst